run llama-3.1-8b on 3 m4 pro cluster

b6f2385c41f1a69924f1d6ad9cab113799b67e8b · 2024-12-12 15:13:10 +0000 · Alex Cheema

Files touched

M .github/workflows/bench_job.yml
M .github/workflows/benchmarks.yml

Diff

commit b6f2385c41f1a69924f1d6ad9cab113799b67e8b
Author: Alex Cheema <alexcheema123@gmail.com>
Date:   Thu Dec 12 15:13:10 2024 +0000

    run llama-3.1-8b on 3 m4 pro cluster
---
 .github/workflows/bench_job.yml  | 3 +++
 .github/workflows/benchmarks.yml | 2 +-
 2 files changed, 4 insertions(+), 1 deletion(-)

diff --git a/.github/workflows/bench_job.yml b/.github/workflows/bench_job.yml
index fe3ea9b3..d40e9fd1 100644
--- a/.github/workflows/bench_job.yml
+++ b/.github/workflows/bench_job.yml
@@ -80,6 +80,9 @@ jobs:
           
           # Source the performance environment variables
           source /tmp/performance_env.sh
+
+          # MLX Memory Settings
+          ./configure_mlx.sh
           
           # Verify optimizations
           echo "Verifying performance settings..."
diff --git a/.github/workflows/benchmarks.yml b/.github/workflows/benchmarks.yml
index 9790414b..e00b5dfd 100644
--- a/.github/workflows/benchmarks.yml
+++ b/.github/workflows/benchmarks.yml
@@ -24,7 +24,7 @@ jobs:
   three-m4-pro-cluster:
     strategy:
       matrix:
-        model: ['llama-3.2-1b', 'llama-3.2-3b']
+        model: ['llama-3.2-1b', 'llama-3.2-3b', 'llama-3.1-8b']
       # Optional: add fail-fast: false if you want all matrix jobs to continue even if one fails
       fail-fast: false
     uses: ./.github/workflows/bench_job.yml

← 9472ab0d t · back to Exo · add llama-3.2-1b-8bit, llama-3.2-3b-8bit, llama-3.2-3b-bf16 a0bada3b →